telega/webhook

What every webhook adapter checks before an update reaches the bot.

telega_wisp and telega_mist are thin: they read the request and answer it. The decisions — is this the bot’s path, does the secret match, is the bot in a state to take an update, is this a health probe — live here, so an adapter for another server is a few lines over admit and health_probe, and all adapters reject and report alike.

Types

Admission

</>

Whether a request on the bot’s webhook path may be processed.

pub type Admission {
  NotWebhook
  Rejected(status: Int)
  Admitted
}

Constructors

```
NotWebhook
```
Not the bot’s webhook path: the adapter’s other routes get the request.
```
Rejected(status: Int)
```
The bot’s path, but not to be processed: answer status and nothing else. 401 when the secret token is wrong; 503 when the bot is draining, over its with_max_in_flight cap or not answering, so Telegram redelivers the update after the deploy or the spike instead of it being lost.
```
Admitted
```
Read the body, decode the update and hand it to the bot.

Values

admit

</>

pub fn admit(
  telega telega: telega.Telega(session, error, dependencies),
  path_segments path_segments: List(String),
  secret_token secret_token: Result(String, Nil),
) -> Admission

Decide what to do with a request, from its path segments and the value of secret_header (Error(Nil) when absent).

The checks run in this order, so a request that is not ours costs nothing and a wrong secret is turned away before the body is read.

default_health_path

</>

pub const default_health_path: String

The path health_probe answers on when an adapter picks the default.

health_probe

</>

pub fn health_probe(
  telega telega: telega.Telega(session, error, dependencies),
  method method: http.Method,
  path_segments path_segments: List(String),
  path path: String,
) -> Result(#(Int, String), Nil)

Ok(#(status, json)) when the request is a health probe — a GET on path (leading and trailing slashes ignored) — and Error(Nil) for any other request.

The status is 200 while the bot actor is alive, accepting updates and below the cap set by telega.with_max_in_flight, 503 otherwise; the body is telega.health_to_json. That is what a load balancer, a Kubernetes readiness probe or a fly.io health check wants: a deploy drains out of rotation instead of black-holing updates.

secret_header

</>

pub const secret_header: String

The header Telegram sends the secret_token of setWebhook in.